Chrome

Trying to do better...

[Matt Pocock @mattpocockuk]

Of all the things I've tried to get good at, teaching has been the most rewarding

Figuring out the best way to compress knowledge and fit it in people's heads in an engaging way is a puzzle that never ends

Every course I teach, I get better, but there is always headroom

[Chrome @ChromeSub]

Here's something really cool that I didn't know Claude could do. I started a conversation and then realized I wanted that to pick up from an earlier conversation in my history. I asked if it could do it.

It found the conversation, summarized it, and brought it into the thread to pick it up from there.

Sweet...

[Noah @NoahKingJr]

I feel like a laptop with 90 tabs open

[Chrome @ChromeSub]

@sudoingX This is the way I came to feel about my photography. It didn't matter if anybody ever saw my photos.

[Chrome @ChromeSub]

So how do I take all of my Adobe Connect recorded sessions (a layout with talking head video, shared screen, animations, and PDFs) to feed this approach?

I have Qwen 3.6 Dense or MOE running on a 3090, but I feel like I would need to use a frontier model to perhaps create code to analyze all those sources to condense to something that could be analyzed on the 3090.

The training materials are static, so it seems to make sense to create some intermediate form. What would that be?

[Ole Lehmann @itsolelehmann]

i'm fully convinced this is the future of education

Matt Pocock built a Claude skill called /teach, and the whole idea is a private tutor that builds an entire customized curriculum around YOU.

think about how school works now.

everyone gets the same lessons at the same pace, whether it's too slow for you or way over your head.

but a great private tutor does the opposite.

they watch where you specifically keep getting stuck, then drill that one weak spot until it clicks.

that's what /teach does automatically.

it finds the bottleneck in your learning and breaks it, over and over, until the thing you couldn't do becomes easy.

he used this skill to learn the Rubik's cube.

it found good sources, wrote him custom lessons with diagrams and little practice drills, and kept a running record of how he was doing.

the way it knows how he's doing is simple: he just tells it. as he practices, he reports back ("i can make the white cross," or "i can mostly solve it but i keep failing the corners"), and it writes that down.

so when he said he was stuck on one specific move, it built the next lesson only for that.

the reason it can do this is memory. most AI forgets everything the moment you close it, so you're always starting from zero.

/teach saves those notes about you on your computer and reads them back before every lesson. so it remembers your goal, what you've already learned, and exactly where you're struggling, then aims the next lesson right at that.

and this works for anything. languages, chess, guitar, onboarding a new hire to a company.

you point it at a topic and it builds you a personal course that keeps adjusting to you and gets smarter the more you use it.

[Matt Pocock @mattpocockuk]

I poured my 10 years of teaching experience into a skill.

It's called /teach, and it can teach you anything.

Here's how it taught me to solve a Rubik's cube:

[Chrome @ChromeSub]

@The_Colony_CC @tonysimons_ According to Claude, the problem is that the API to get the local models doesn't include the context size currently consigured. The openrouter api includes it, supposedly.

[Chrome @ChromeSub]

@justimportant2 @Polymarket 😉

[Chrome @ChromeSub]

@frank_rosh @NVIDIARTXSpark For the early versions, they probably feel like if you even ask, you can't afford it. I will always ask...

[Lawrence McDonald @Convertbond]

Seriously, why even buy an IPO after venture capital has picked over the equity for a decade??? Just buy the post-lockup dump-athon.

h/t Truist's Keith Lerner

cc @jimcramer @andrewrsorkin @BeckyQuick @SullyCNBC

[Chrome @ChromeSub]

@Web3Twon I planned to put 128 in my 2024 build, but people were saying the motherboards weren't stable with four sticks. I went 64. Now the firmware updates fixed it, but too expensive.

[Chrome @ChromeSub]

@HermesAgentTips Nice... it is noticeable. Thanks!

[Chrome @ChromeSub]

@catchvartan @MostlyMonkey If you have 60k SS and all of your big ticket items are paid off, like car and home, it shouldn't take more than 2 million, if you live in the South. No idea how expensive it can be to live in major cities or high cost states.

[OnlyCFO @OnlyCFO]

CFOs reviewing the latest Anthropic bill

[Hasan Toor @hasantoxr]

This open-source app turns Hermes Agent into something normal people can actually use.

It’s called Hermes Desktop.

Instead of managing an AI agent through the terminal, it gives you a real desktop interface for setup, chat, memory, tools, providers, and scheduled tasks.

Basically:

Hermes Agent = powerful brain
Hermes Desktop = control room

What it gives you:

• One place to install Hermes
• Provider setup
• Chat sessions
• Profile switching
• Memory management
• Skills
• Tools
• Scheduling
• Messaging gateways
• Logs
• Backups

Most AI agents are impressive on GitHub but annoying in real life.

They hide state in files.
Break quietly.
Force you into terminals.
Make simple things feel technical.

Hermes Desktop fixes the interface problem.

This is what open-source agents need more of.

GitHub: https://t.co/iHR9kQNJq1

[Michael Guo @Michaelzsguo]

Found this great tool that may be handy for your local LLM inference optimization:

https://t.co/BqX3mZJEhU

And apparently 1M tokens for DeepSeek V4 Pro only takes 5GB of RAM.

What the heck?

[Chrome @ChromeSub]

@Hikari_07_jp Many years ago, while traveling for work, I always thought it would be awesome if you could get on the plane and know that you were sitting down next to someone who had a similar interest so you could have a great discussion.

Well... X is that seat....

[Chrome @ChromeSub]

@Haryormdey4u And take out his hearing aids to recharge....

[Chrome @ChromeSub]

@KuittinenPetri @svpino If you had to choose only one, would it be the AMD or NVIDIA? I have a 3090 in a W11 PC that is fine for fast, smaller models. I'd like to supplement it for running larger models.